Identification and Characterization of Vena Cava Filters
Purpose | Tool to automatically identify and characterize (type, location, potential complications) vena cava filters on imaging studies. |
Tag(s) | |
Panel | Interventional Radiology |
Define-AI ID | 19090001 |
Originator | Jeremy C. Durack, MD |
Lead | Jeremy C. Durack, MD |
Panel Chair | Jeremy C. Durack, MD; Rajesh Shah, MD |
Panel Reviewers | Interventional Radiology Panel |
License | |
Status | |
RadElement Set | RDES104 |
Clinical Implementation
Value Proposition
The majority of vena cava filters placed today are retrievable, though it is known that filters are often not retrieved even when clinically appropriate. Identification and documentation of filter presence could trigger inclusion on a patient's problem list to encourage discussion regarding appropriate and timely retrieval. In addition, complications such as migration, fracture, thrombosis, and vena cava perforation are well known and detection may be facilitated using using machine learning techniques.
Narrative(s)
A patient with a history of deep venous thrombosis had a filter placed pre-operatively. The patient recovered from the procedure and has restarted anticoagulation. The mechanical filter is no longer required, however due to transitions of care teams, no one has taken responsibility for filter management. This tool would help to identify the presence of a filter and that information could be used to trigger appropriate management discussions as well as alert clinicians to potential filter-related complications.
Workflow Description
Abdominal x-ray or CT imaging of the abdomen performed. Automated detection of a filter, matched against training set of filters to determine type and location relative to the renal veins/vertebral bodies.
Considerations for Dataset Development
Procedure(s) | X-ray abdomen, CT abdomen |
View(s) | All |
Sext at Birth | Male, Female |
Age | All |
Indications for filter | varied |
Filter indwelling time | varied |
Filter Location | varied (see locations below in technical specifications) |
Filter type | varied (see types below in technical specifications) |
Filter injection route | varied |
Technical Specifications
Inputs
DICOM Study
Procedure | X-ray, CT abdomen |
Views | All |
Data Type | DICOM |
Modality | X-ray, CT |
Body Region | Abdomen |
Anatomic Focus | Vena cava |
Pharmaceutical | |
Scenario | MRI may have limited utility in filter assessment due to metallic artifcacts |
Primary Outputs
Filter Present
RadElement ID | RDE649 |
Definition | Presence of filter(s) |
Data Type | Categorical |
Value Set |
|
Units | N/A |
Filter Component count
RadElement ID | RDE646 |
Definition | Count of filter components |
Data Type | Numeric |
Value Set | N/A |
Units | Number of components |
Filter Type
RadElement ID | RDE651 |
Definition | For each filter identified, return the type of filter |
Data Type | Categorical |
Value Set |
|
Units | N/A |
Filter or filter component location(s)
RadElement ID | RDE652 |
Definition | Anatomic location of filter(s) or filter component(s) |
Data Type | Categorical (multi-select) |
Value Set |
|
Units | N/A |
Secondary Outputs
Filter Complications
RadElement ID | RDE653 |
Definition | Radiographically evident filter complications |
Data Type | Categorical (multi-select) |
Value Set |
|
Units | N/A |
Future Development Ideas
Include features like:
-
Change of filter component location over time
-
Distinguish features of multiple filters or filter components present